Sam Altman’s OpenAI comeback is a strategic win for Satya Nadella

Committing $11 billion that gave Microsoft a 49% stake — and no board seat was an extraordinary gamble by Nadella, given the possibility of Altman being fired.

Of all the stories of a hero’s exile and return from the wilderness, none has caused more whiplash than Sam Altman’s. It took Steve Jobs a decade to return to Apple Inc. after being knifed by his board; Jack Dorsey took seven years to come back to Twitter, and an entire adolescence and early adulthood passed for Simba to be crowned the Lion King. Altman’s round trip as chief executive officer at OpenAI took fewer than five days. All of which is very, very good for Microsoft Corp.

After his shock termination last Friday, Altman is returning to a startup with a clearer corporate direction. OpenAI’s board, previously made up of three staff members and three independent directors with ties to the effective altruism movement, and which had a fiduciary duty to humanity rather than investors, will now look much more like a typical tech company board.

It’ll have nine seats, at least one of which will likely go to Microsoft (at minimum as an observer). And the confirmed board members scream “safe pair of hands” to OpenAI’s investors and customers. They include Bret Taylor, former co-CEO of Salesforce Inc. and the adult-in-the-room on Twitter’s board when Elon Musk scooped up the company, and Larry Summers, the former US Treasury Secretary and a paid contributor to Bloomberg Television. The two have served on an array of corporate boards and are well-versed in serving the needs of investors like Microsoft.

OpenAI has yet to confirm what will happen to the three independent board members who voted Altman out. The New York Times reported Wednesday that OpenAI’s chief scientist, Ilya Sutskever, academic Helen Toner and robotics entrepreneur Tasha McCauley have all agreed to step down. Quora CEO Adam D’Angelo, who was among the four board members to vote out Altman, appears for now to be staying.

MC Explains: What is AGI? and what does it have to do with OpenAI?

As Bloomberg Opinion columnist Matt Levine outlined Tuesday, OpenAI’s efforts to thread the needle between governing a for-profit company and a nonprofit organization were convoluted and doomed for its investors from the start. At the very top of OpenAI’s byzantine structure was a board that answered only to their moral instincts. For some still-inscrutable reason, they deemed Altman dishonest, and thus an impediment to the safe harboring of AGI (artificial general intelligence), or AI systems that could surpass human intelligence.

Although the setup was quite good for humanity, it was always a problem for Microsoft. It’s why Satya Nadella and his chief financial officer, Amy Hood, initially hesitated to invest $1 billion in OpenAI in 2019, according to a person familiar with Nadella’s thinking. But they eventually committed the money that year, and then another $10 billion last January to give them a 49% stake — and no board seat.

That was an extraordinary gamble by Nadella. He knew from the beginning that the board could fire Altman at any point. Altman even reiterated that possibility in interviews, saying it was a good thing that they could boot him out. It made him accountable to humanity’s future.

Also Read: Despite AI taking over, demand for animation talent holds on to its sheen

Now he will be more accountable to investors like Microsoft and Nadella, who says “surprises are bad” and who definitely won’t be blindsided again by the actions of AI safety advocates. The new board will probably neuter OpenAI’s nonprofit arm so it no longer hangs over the company like the Sword of Damocles.

In case you missed it over the weekend of bizarre twists and turns, Microsoft briefly hired Altman and his OpenAI President Greg Brockman as heads of a new advanced AI division at Microsoft. As nearly all of OpenAI’s staff signed an open letter threatening to join that team, many observers in the industry saw an enormous coup for Microsoft: It had acqui-hired one of the world’s most talented AI teams for approximately zero dollars.

But that isn’t quite right. OpenAI’s staff wouldn’t have come free since there were 770 of them and many had seven-figure salaries. And more importantly, their work would have loaded Microsoft with more risk. Microsoft’s partnership with OpenAI worked so well until now because OpenAI took all the reputational and legal flak for deploying advanced AI systems like ChatGPT and DALL-E 2 into the world.

That was why they partnered in the first place, and it’s why tech giants like Alphabet Inc.’s Google and Microsoft have been so slow on AI development themselves. Remember Microsoft’s PR disaster with Tay, the chatbot that became a white supremacist on Twitter? Microsoft employees still cringe at the scandal, and it made others wary of releasing their own large-language model technology.

But OpenAI is a startup and can get away with throwing its cutting-edge AI experiments into the wild. Doing so generates feedback and hype, which rubs off nicely on Microsoft’s all-important Azure cloud platform.

This may interest you: The winners and losers in Sam Altman’s OpenAI kingdom

By returning to its hands-off partnership, Microsoft gets all of the glow and none of the liability, plus more control. That is a much better deal than having OpenAI in-house, its young futurists grappling with the bureaucracy of a tech giant where everyone wears khaki pants and has been wandering around the office for decades.

Nadella can silently thank OpenAI’s nonprofit board for causing all this ruckus. The outcome isn’t even a silver lining — it’s a stronger position in the AI arms race.